political entry, public policies, and the economyone reason that political institutions a⁄ect...
TRANSCRIPT
NBER WORKING PAPER SERIES
POLITICAL ENTRY, PUBLIC POLICIES, AND THE ECONOMY
Casey B. MulliganKevin K. Tsui
Working Paper 13830http://www.nber.org/papers/w13830
NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts Avenue
Cambridge, MA 02138March 2008
We appreciate the comments of Gary Becker, Bill Dougan, Roger Myerson, Jose Plehn-Dujowich,John Sutton, Chad Syverson, Sven Wilson, workshop participants at Buffalo, Clemson, Harvard, andThe University of Chicago, and the financial support of the University of Chicago's Stigler Centerfor the Study of the Economy and the State. The views expressed herein are those of the author(s)and do not necessarily reflect the views of the National Bureau of Economic Research.
NBER working papers are circulated for discussion and comment purposes. They have not been peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies officialNBER publications.
© 2008 by Casey B. Mulligan and Kevin K. Tsui. All rights reserved. Short sections of text, not toexceed two paragraphs, may be quoted without explicit permission provided that full credit, including© notice, is given to the source.
Political Entry, Public Policies, and the EconomyCasey B. Mulligan and Kevin K. TsuiNBER Working Paper No. 13830March 2008JEL No. H11,L12,P16
ABSTRACT
This paper presents a theory of competition for political leadership between incumbent leaders andtheir challengers in which the possible equilibrium political market structures range from pure monopoly(unchallenged dictatorship) to perfectly competitive (ideal democracy). Leaders are constrained bythe threat of "entry" or their ability to tax (or both), so that regimes with no challengers may nonethelessimplement policies in the public interest. We offer economic interpretations of why democratic countriesare associated with higher wages, why resource abundant countries tend to be nondemocratic, andhow technological change affects political development. By focusing on the incentives for politicalentry, we show how trade sanctions and other policies designed to promote democracy may actuallyhave the unintended consequences of discouraging political competition.
Casey B. MulliganUniversity of ChicagoDepartment of Economics1126 East 59th Street, #506Chicago, IL 60637and [email protected]
Kevin K. TsuiClemson UniversitySirrine HallClemson, SC [email protected]
I. IntroductionPolitical entry is likely a¤ected by the economic conditions, and has an in�uence on public policy.
Our goal is to specify some of the basic incentives for political entry that would be common across
a variety of institutional and economic environments. In doing so, we derive e¤ects of economic
development, technical change, and other characteristics of the economy on political entry, as well
as e¤ects of political entry on public policy. Not surprisingly, citizens are better o¤ when the
political sector is competitive: that is, when their incumbent leaders make policy decisions with a
sensitivity to the threat of political challenges. However, we show that political market structure,
measured by the number of actual challengers, can be a poor indicator of political competitiveness.
Rather, political competitiveness is better measured by the size of political entry barriers.
Political entry barriers are widely discussed in political science, with a large section of com-
parative political science concerned with measuring them. Observation suggests that there may
be no ideal democracy on earth, in the sense that anyone can costlessly enter the competition
for public o¢ ce. At the same time, few (if any) polities are fully monopolized, because even the
most oppressive regimes show some sensitivity to popular support, and some concern that a lack
of popular support would hurt the regime�s survival and e¤ectiveness.1 However, many formal
models in economics and political science do not consider entry barriers, or at most treat them as
a �xed parameter.2 We help �ll this gap by building a model of the causes and consequences of a
variety of autocracies and democracies, identi�ed with di¤erent degrees of entry barriers and hence
competitiveness.
With a model of political entry, we are able to o¤er economic interpretations of how dictators
extract more political rent compared with democratic leaders, why democratic countries are as-
sociated with higher wages, why resource abundant countries tend to be nondemocratic, and how
technological change a¤ects political development. Interestingly, political competitiveness may have
little e¤ect on a wide range of economic and social policies, like the mix of taxes, or spending on
Social Security. Instead, competitiveness is re�ected by policies like military spending, torture, and
execution, which more directly serve to protect the incumbent leader�s position. By focusing on the
incentives for political entry, it is also easy to see how trade sanctions and other policies designed
to promote democracy may actually have the unintended consequences of discouraging political
competition.
Unlike a typical �rm that shares a market with his competitors in an industry, government
1According to the POLITY IV (2000) indices of political competitiveness, at least 80% of the world, and 94%of the nonOECD countries, are imperfectly competitive in terms of the selection of political leaders, or in termsof the degree to which alternative views on policy and leadership can be expressed in the political arena. Moreimportantly, the indices exhibit signi�cant variations across countries and over time, suggesting that it is importantto treat political competitiveness as an endogenous variable.
2Earlier models usually assumed either perfect competition or no competition of any kind. Perfect competitionis typically assumed in models in which public policies are determined by universal voting or economic e¢ ciency(Becker, 1958; Stigler, 1972; Wittman, 1989, 1995). The monopoly models of government include Breton (1974),Brennan and Buchanan (1980), and Olson (1993). An exception is Grossman (1991), although the paper focuses onmass insurrections instead of entry from other potential political leaders. An overview of the more recent literaturewill be provided in section II.
1
has long been understood as a natural monopoly on force. However, Schumpeter (1942), Becker
(1958), and Tullock (1965) explained how government may only be a monopoly in a static sense,
and an ideal democracy regularly has perfect competition for the right to run the monopoly until
the next election. It is our contention that this logic can be generalized to all regimes, even when
regime turnover may be driven by revolutions or insurrections. In this regard, both democratic
and nondemocratic government have something in common with a regulated public utility. For
example, only one �rm at a time can deliver electricity, but the �rm doing so may compete with
others for the job, perhaps via a license auction, by pleasing a regulator who answers to the voters,
etc. We interpret the degree of political (non)competitiveness as the size of entry barriers into the
process allocating the rights to temporarily run the government, or the natural monopoly on force.
Such a dynamic political competition also resembles a sequential patent race (Reinganum,1985),
in which �rms exert R&D e¤ort in innovation competition and a successful �rm can enjoy the tem-
porary monopoly power granted by the patent until he is �overthrown�by another more inventive
challenger. After a review of previous studies from the related literature, our section III begins
with a simple model of imperfect political competition, akin to a patent race.3 Perhaps one ma-
jor di¤erence between a dictator and the manager of a public utility is that the former has no
higher government to enforce agreements between he and his �customers.� Hence, following the
modern political economy literature, we assume that leaders lack the ability to commit to their
policy platform until they are in power. More importantly, we generalize the existing patent mod-
els by endogenizing the size of entry barriers, which is arguably more important in the political
sector. As a result, the political market can have various structures ranging from the most repres-
sive leviathan to a perfect democracy. In all regimes, competition creates a fundamental con�ict
between incumbent leaders and challengers.
Rents are created and limited by entry barriers. Section IV demonstrates the complementarity
between rent extraction and entry barrier policies, so that a more repressive regime will extract
more rent. Our comparative statics predict that countries with higher wages tend to have a more
competitive political sector, measured either by lower entry barriers or rent extraction. Depending
on the source of income, however, economic development may or may not encourage political
competition. By emphasizing the role of entry barriers, our theory also provides a novel prediction
on democratization (i.e. erosion of political entry barriers) based on enforcement technological
change.
Section V investigates some consequences of political competitiveness. First, we identify condi-
tions under which all regimes choose the public policies that maximize the welfare of the society,
conditional on the citizens� income. Because entry barriers are costly to maintain, leaders from
regimes that are threatened by entry care about popular support. Moreover, when actual or even
potential political competition is su¢ ciently intense so that rent extraction is limited purely by
3Wittman (1989, 1995) adapted private sector competitive theory to the public sector in order to formulate atheory of (ideal) democratic performance. Our paper pushes Wittman�s private-public analogy beyond the purelycompetitive case. See also Baye and Hoppe (2003) for formal analogies between rent seeking contests and patent-racegames.
2
threat of entry but not technology of taxation, both democracies and nondemocracies will choose
the same public policy. Similarly, when leaders are constrained by competition instead of their
ability to tax, we show that the actual and nominal incidence of foreign policies directed at �mo-
nopolized�political leaders are di¤erent, because of their impact on supply conditions.4
II. Previous Studies of Entry or Monopoly in the Political Sector and Their E¤ectson Public Policy
A few previous papers have modeled candidate entry.5 Some of them (such as Besley and
Coate, 1998) have no entry barriers, and �nd equilibrium rents to be zero. Besley and Coate (1997)
predict that uncontested elections are more likely when entry barriers are high. Fedderson, Sened,
and Wright (1990) formulate a model with positive entry barriers in order to show that equilibrium
candidates are still likely to be centrist even when there are more than two of them. Myerson (1993)
and Persson and Tabellini (1999, 2000, Ch. 9) compare two electoral systems, and �nd that the
system with lower entry barriers has more political parties and smaller rents for o¢ ce-holders. The
incumbency advantage has been attributed to the existence of entry barriers, and the well-known
Duverger�s law, which says the plurality rule election system tends to favor a stable two-party
system while a proportional representation system fosters party development, can be interpreted
as plurality voting�s creating a barrier to entry against new third parties (Myerson, 1999).
The aforementioned models of candidate entry feature a simultaneous game with exogenous
political survival: candidates are competing against each other, and not against the incumbent.
We model a sequential game in which entry barriers can favor incumbents by limiting competition.
The decision to compete depends, of course, on the size of the entry barrier, but also on what the
incumbent is doing. Thus, in our model an incumbent can discourage entry by behaving well.6
One reason that political institutions a¤ect public policies is that they a¤ect the identity of the
pivotal voter.7 Another reason is that policies depart from the pivotal voter�s preferred policies
(such as rents going to o¢ ce-holders), and institutions a¤ect the magnitude and direction of such
departures. Our model abstracts from the e¤ects of voting rules and other institutions on the
identity of pivotal voters by having a representative voter/citizen.
Several studies have helped explain why politicians might receive rents at taxpayer expense.
Some of these (e.g., Polo, 1998) rely on imperfect commitment by electoral candidates to the
policies they will implement when elected, in which case o¢ ce-holders are compensated in order
4Related incidence results for the private sector are familiar from the industrial organization and public �nanceliteratures.
5Many papers have also looked at practical examples of political entry barriers, including Tullock (1965), Crain(1977), Lott (1986), Gelman and King (1990), Friedman and Wittman (1995), and Wohlgemuth (1999), although notnecessarily connecting them to public policy outcomes.
6This e¤ect of competition, including potential competition, is well-known in the (private sector) industrial orga-nization literature (e.g. Davis, Murphy, and Topel, 2004; Goolsbee and Syverson, 2006). Grossman and Noh (1990,1994) apply this idea to the political sector; however, they do not model entry and entry barriers explicitly.
7For example, majoritarian (or multiple-district) elections may induce politicians to pay more attention to votersin marginal electoral districts, whereas proportional (or single-district) elections induce parties to seek support frombroader coalitions in the population. See Persson, Roland, and Tabellini (2000) and Lizzeri and Persico (2001) forstudies of how speci�c constitutional rules can a¤ect public policy, and Persson and Tabellini (2000, Ch. 8) for asurvey of the literature.
3
to correctly reveal the state of nature and implement appropriate public policies.8 Others, such
as probabilistic voting models (e.g. Persson and Tabellini, 1999), assume that candidates supply
something in addition to public policies (such as ideology, personal appearance, etc.). In both types
of models, rents are possible because the number of candidates is �xed and �nite.9 Presumably
entry barriers are required to so limit the number of candidates, although we are not aware of a
formal political model of this type.
The existing literature on equilibrium rents with a �xed number of challengers sometimes �nds
that public policies coincide with those preferred by the pivotal voter, aside from the excessive
taxation required to �nance the rents for o¢ ce-holders.10 With an endogenous number of chal-
lengers, we �nd that some, but not all, public policies are invariant to the magnitude of political
entry barriers and the number of challengers, and thus would be the same in otherwise similar
democratic and nondemocratic countries.
Autocracy has been alternatively viewed as (a) the no-entry outcome in a sequential struggles
among competing political leaders, as in our model, Tullock (1987), and Grossman and Noh (1990,
1994),11 or (b) as the elite-as-victor outcome of a struggles between competing citizen groups, as
in Acemoglu and Robinson (2000, 2001 and 2006), Bueno de Mesquita et al (2003), and Huang
(2007).12 In particular, the latter approach suggests that democracy indicates a sharing of political
power with the poor, and thus political and economic victory for them, whereas the former approach
predicts that autocrats will help the poor for the same reasons that democratic leaders do.
In our model, democracies di¤er from nondemocracies in a continuous fashion: in terms of the
relative in�uence of incumbents and challengers.13 Several studies have modeled one of the two
extremes: democracies as competitive suppliers of public sector goods and services and autocracies
that are monopolists of the public sector, extracting the maximum possible revenue from their
�customers� (e.g. North, 1981).14 Olson, McGuire, and Niskanen predict that dictators consume
all public funds for themselves (except perhaps for expenditures to enhance the tax base) so that,
among other things, dictators do nothing to help the poor.15 In this regard, they appear to agree
with Acemoglu, Robsinson, Bueno de Mesquita et al., that democracies are better for the poor.
8Our model does not feature this e¤ect � it has no commitment yet zero rents would still be the outcome (withzero entry barriers) because citizens in our model are not trying to incentivize o¢ ce-holders.
9Polo (1998, Proposition 9) shows how more candidates dissipate political rents in his imperfect commitmentmodel. Persson and Tabellini (2000, p. 73) conjecture that more candidates would dissipate political rents inprobabilistic voting models, and that equilibrium rents are implicitly the consequence of entry barriers.10See Persson and Tabellini (2000, p. 72).11See also Gallego and Pitchik (2004) on dictator turnover, and Wintrobe (1990) on how dictators combine loyalty
and repression in order to survive.12Political and economic historians have documented that revolutionary challenges to elites come most often from
competing elites. In a sample of 89 heads of authoritarian government that held o¢ ce for at least one year and lostpower by irregular means between 1950 and 1990, Svolik (2006) �nds that only 6 lost power from because of popularuprising, while the remaining were removed by other political competitors.13Schumpeter (1942), Becker (1958), and Tullock (1965) also classify polities by the degree of competitiveness,
albeit informally.14Brennan and Buchanan (1980) have a model in which even democracies are leviathans, except as limited by their
constitutions. See Wintrobe (2000, p. 131) for a list of several other studies modeling dictators as monopolists of thepublic sector.15Olson (1993), Olson and McGuire (1996), and Niskanen (1997).
4
Wintrobe (2000) also considers autocracies to be monopolists, but comes to a di¤erent conclusion.
In his view, an interest group or voting block that is powerful under democracy is even more
powerful under dictatorship because they can repress their political opponents. If the typical
democracy redistributes from rich to poor, then Wintrobe predicts that the typical nondemocracy
will redistribute even more from rich to poor.16
III. A Model of Sequential Political CompetitionWe present in this section a model of political leadership as a monopolistically competitive
patent race with endogenous entry barriers.
III.A. Three Types of Public Policies and the Tax ConstraintPublic policies are of interest to the government leadership because they a¤ect longevity and
leadership income. We partition public policies into three types: (1) �social and economic�policies
x; (2) �barriers to entry�policies b; and (3) rent extraction r:17
Social and economic policies like provision of public goods, social security, the minimum wage,
various rates of taxation, etc., are functionally unrelated to the blocking of political challengers.
These policies do not a¤ect political competition, except indirectly by enhancing the government�s
popular support.
Barriers to entry policy instruments, like execution, torture, the degree of censorship, the orga-
nization of the military, ballot fees, vote quotas, etc., however, have the primary e¤ect of blocking
political competition. Moreover, b is not a money transfer from the challengers to the incumbent
and thus is a pure deadweight loss. As with the study of complements and substitutes in consumer
and producer theory, the distinction between x and b should be made on �rst principles, so that
theoretical implications for x-b co-movements are testable rather than tautological. Some of these
judgements are straightforward, as it seems that censorship and torture have a much di¤erent func-
tional relationship with blocking political challengers than does, say the minimum wage rate. Even
if these judgements turn out to be di¢ cult, the assumption that there are a lot of important policies
in the x vector is enough to generate some interesting results, namely that democracy should not
a¤ect public conduct in many dimensions.18 Although in application it is useful to interpret x and
b as vectors of policies, for simplicity we henceforth assume that x and b are each scalars.
Market power permits leaders to in�uence public policies for their satisfaction or personal pro�t,
16Wintrobe�s approach is also unique in that it does not hinge on the assumption that redistribution is necessarilyfrom rich to poor. Thus, Wintrobe�s more general result is that dictatorships redistribute in the same direction thatdemocracies do (e.g., from young to old, or from consumers to producers), but in greater magnitude.17 In the literature on constitution, policies are partitioned into those that can provide bene�ts to (1) many citizens,
such as broad programs in the form of general public goods, (2) a narrow group of citizens, such as barrel-porkprojects, or (3) virtually no citizens, but a speci�c group of politicians, such as salaries for public o¢ cials or variousforms of corruption (Persson and Tabellini 2003, p. 14). We do not explicitly distinguish these �rst two types ofpolicies and we combine them into our policies x: Instead, our model emphasizes two distinctive classes of publicpolicies that facilitate leaders�entry prevention and rent extraction.18 Interestingly, Tullock�s (1987) book � a book about the measures autocrats take to protect o¢ ce � is clear that
some public policies are more important than others when it comes to blocking political challengers: the death penalty(p. 6, 20, 65, 80), torture (p. 61, 62, 64, 65), press freedom (p. 154), regulation of religion (p. 108), and maintainingan army. A whole range of public policies, like education spending, revenue, pension spending, nonpension �social�spending, the corporate tax rate, and payroll taxation are conspicuously absent from his analysis of entry barriers.
5
which is legal if the leader is su¢ ciently convincing as to the public�s interest in the policy. In many
cases, part of the political rent is spent on entry barrier maintenance. In particular, the �ow of
the net political rent is r � �b; where �b is the cost of maintaining political entry barriers in theamount b: We assume that � > 0 and b � 0: In other words, we rule out the possibility that theincumbent can subsidize entry.
The social and economic policy x is associated with a per unit cost p > 0 of policy imple-
mentation, which includes any deadweight cost that reduces national income. Similarly, when x
represents provision of a public good that enhances GDP, p can be interpreted as the net cost of
production. For simplicity, policy x and rent extraction r are �nanced through a lump sum tax
T:19 The tax T itself is a policy; however, given p and the government budget equation, choosing
x and r is equivalent to choosing x and T: The government budget constraint is:
T = r + px � �y
where � 2 (0; 1] is a parameter that denotes the maximum fraction of income y > 0 that is feasibly
taxed. �y can therefore be interpreted as the maximum point of the La¤er curve faced by the
government. y, � ; and p are exogenous parameters in our model.
III.B. Supply Side Determinants of Political Survival: The Entry ConstraintA large pool of identical citizens are potential challengers to the incumbent�s political leadership.
An actual challenger spends resources b attempting to assemble a winning coalition. From the
challengers�point of view, b is a political entry barrier created by the incumbent regime to discourage
their entry; b includes the punishments (actual or potential), censorship, and inconveniences created
by the incumbent in order to block challengers.
Each challenger proposes a policy, which both determines the pro�ts he would earn if successful
and the amount of popular support he enjoys. We assume that challengers cannot commit to their
policy before takeover actually happens:20
Assumption 1. (No Policy Commitment) Each challenger lacks the commitment to set policiesuntil he is in power.
The incumbent is, of course, concerned with the number and actions of his challengers. Each
challenger is concerned with the number, and actions of, other challengers. More important, each
challenger is challenging because he hopes to take over, and himself someday be the incumbent.
Hence, he is concerned with the next generation of challengers, whose probabilities of success will
be determined by their policies relative to his. In other words, by implementing a popular policy,
19For simplicity, we model rent extraction as if it were taken in cash. r is more than corruption, which is sometimesinterpreted as illegal cash receipts by political leaders. Some leaders receive their cash legally, as with monarchswith the legal rights to sell monopoly licenses, or dictators who legally pay themselves large salaries or build palaces,summer homes, etc.20An alternative approach would have challengers commit to their policy platform upon entry into the political
competition. We pursued this approach in an earlier version of this paper. Interestingly, both approaches havesimilar steady state qualitative comparative statics. Our no-committment approach is more tractable and perhapsmore realistic.
6
a leader in power lengthens the expected lifetime of his regime.
Regimes are indexed t = 0; 1; ::: with 0 denoting the incumbent regime. When variables vary
over time, we use subscripts to distinguish one regime�s variables from another�s. For simplicity,
policies are assumed to be constant over time within a regime. For example, bt denotes the entry
barrier protecting regime t:
Let�s consider the situation in which the number of parties challenging regime t is ct: These
challengers are indexed j = 1; 2; :::; ct and have success hazards h1;t; h2;t; :::: Challenger j�s success
hazard hj;t depends on his policies fxj;t+1; bj;t+1; rj;t+1g and the incumbent�s policies fxt; bt; rtg;but for the moment our notation suppresses this dependence until we consider the implications of
maximizing behavior. Regime t�s (�ow) payo¤ to governing is rt � �bt. When the �rst challengersucceeds, the incumbent stops receiving this �ow: If the �rst challenger succeeds at date R in the
incumbent�s regime, the incumbent�s value of governing (from the perspective of the time he began)
is vt(R):
vt(R) = (rt � �bt)Z R
0e�isds = (rt � �bt)
1� e�iRi
;
where i > 0 is the usual interest rate.
The probability that regime t lasts exactly R units of time is cthte�cthtR; where ht is the average
success hazard among the challengers and ctht is the aggregate success hazard. In order to calculate
the expected value of governing as regime t; we integrate vt(R); weighting by these probabilities.
Vt �Z 1
0cthte
�cthtRvt(R)dR =rt � �bti+ ctht
: (1)
In other words, the leadership has a discount rate that combines the usual interest rate with a
hazard rate ctht for ending the regime.
Challenger j�s expected pro�t �j;t from challenging regime t depends on four things: (a) the
hazard of succeeding hj;t; (b) hazard of the incumbent�s falling ctht (where ht denotes the average
success hazard among the ct challengers); (c) aggregate value Vt+1 of beginning the next regime,
and (d) the entry barrier bt: We model b as a one-time cost for a challenger, which gives him a
positive success hazard at each moment in time during which the existing incumbent is still in
power.21
If and when challenger j overthrows regime t, the challenger obtains expected leadership value
Vt+1. This expected prize Vt+1 is discounted by challenger j using the factor hj;t=(i+ ctht) because
1=(ctht) units of time are expected to elapse before regime t falls and, when regime t does fall,
challenger j�s probability of success (rather than one of the other challengers) is hj;t=(ctht). At the
21We have considered two possible variations on this speci�cation. One of them also has a �ow cost of challenging(see Lee and Wilde, 1980) for a private sector patent race model with both stock and �ow costs of challenging), whichmakes the numerator (hj;tVt+1 � bf ), where bf is a �ow cost of challenging.A second variation has the stock entry cost b enabling a challenger to challenge for a �nite period: This version has
the same qualitative implications as our present model (it modi�es only the pro�t function by changing the functionalform by which r + ch enters), so for simplicity we do not present them.
7
point of making his entry decision, challenger j�s expected pro�ts:
�j;t =hj;tVt+1i+ ctht
� bt:
More challengers lower the expected pro�t from challenging. Let w be the opportunity cost
of challenging for any potential challenger. In other words, w � 0 is the reservation wage of
challengers.22 Challenger j to regime t earns zero expected pro�ts when
hj;tVt+1i+ ctht
� bt = w: (2)
While bt is an endogenous entry barrier chosen by the incumbent leader, w can be interpreted as an
exogenous entry barrier determined by the condition of the economy. When citizens�productivity
is high in the private sector, other things being equal, the incumbent faces a smaller threat of entry
because the opportunity cost of challenging is high.
The entry constraint for the incumbent (2) is therefore also a zero pro�t condition for actual
challengers j = 1:::ct. If the number of challengers is continuous and potential challengers j = ct:::1are atomistic,23 then equation (2) is also a zero pro�t condition for potential challengers. The
above entry condition will hold in equality in equilibrium as long as there is entry and the number
of challengers is endogenously determined. However, our model does not rule out the possibility
of zero entry because the opportunity cost of entry is prohibitive. In this corner-solution of zero-
challenger equilibrium, we havehj;tVt+1i
� bt � w:
III.C. Demand Side Determinants of Political Survival: Popular SupportThe incentives for public policy choices are related to the form of the hazard function h. We
suppose that a leader�s decisions can be in�uenced by �popular support�even if he is a dictator.
We model a leader�s popular support S as a function of utility �ow of a �representative�citizen:
S = u(z; x) (3)
where the utility function u represents a policy preferences of the representative citizen, and z
denotes his income.24
22A citizen faces a choice between being a political entrepreneur or working in the private sector. This is similar toLucas�s (1978) model of career choice between an entrepreneur or an employee in a private economy, where the marketwage rate measures the opportunity cost of the entrepreneur�s time. The general point that political participationhas a private sector opportunity cost is also made in Mulligan and Sala-i-Martin�s (1999) analysis of social security.23These conditions would hold, for example, if the number of potential challengers were large and the strategy
space for challengers and potential challengers were a probability of entry. In this case, ct would be interpreted asthe expected number of challengers, with the actual number to be determined by the random mechanism associatedwith the mixed strategy (see, e.g. Dixit and Shapiro, 1986). We follow much of the literature on private sector entryand ignore the constraint that ct must be an integer.24u can be interpreted as an indirect utility function embedding household maximization behavior subject to policy
8
The public sector changes income of the representative citizen. Taxation reduces, and the
social and economic policy x may increase or decrease the income of the representative citizen. The
representative citizen�s disposable income is
z = y � T = y � r � px:
Under some conditions, social and economic policies x are e¢ cient in the sense that the citizens�
marginal rate of substitution between private spending and public policy is equal to the marginal
social cost of that policy, p. Since this outcome is a possibility, it is useful to have some notation
that represents e¢ cient outcomes.
De�nition 1. (Demand for Social and Economic Policies) Let x�(y�r; p) = argmaxxu(y�r�px; x)denote citizens�demand for public policies x and let u�(y � r; p) = u(y � r � px�; x�) denote theutility attained when public policies are provided in the amount demanded by the citizens.
To derive the comparative statics, we assume the following.
Assumption 2. (Citizens�Preferences) (a) u�(z;p)u�z(z;p)
is increasing in z; (b) u(z;x)ux(z;x)
is increasing in x;
and (c) z and x are normal goods.
Note that assumption 2(a) is indeed weaker than concavity of the indirect utility function,25 and
assumption (b) is a regularity condition which guarantees the second-order condition is satis�ed
when the tax constraint is binding.26 As we will see, assumption 2(c) is assumed so that tax revenue
is increasing in rent extraction; many of the comparative statics do not require it.
The identity of the representative citizen can be interpreted in several ways. First, in a ho-
mogenous society, citizens are identical and hence any of them represents the true preferences of
the society. Second, the representative citizen can be interpreted as a pivotal voter/citizen, such
as the median voter, or hypothetical person whose utility function coincides with a weighted social
welfare function that aggregates the preferences in a heterogenous society. Political in�uence in this
case may well be distributed unequally across citizens. The probabilistic voting literature provides
both positive and normative microfoundations of the existence of the representative citizen as the
pivotal voter (Coughlin, 1992). We contend that the same logic applies to all regimes as long as
the threat of entry constrains leaders�behavior and hence leaders care about support from citizens,
even in the absence of meaningful elections under nondemocracies. To focus on the direct e¤ects of
barriers to entry and rent extraction policies and to contrast the existing literature that emphasizes
the con�ict between leaders and citizens, we abstract from any dependence of the identity of this
pivotal voter on political regime, measured by b. i.e. u does not depend on b because there is a
stable distribution of political power among citizens across regimes.
Equation (3) is de�ned in absolute terms. Obviously, support matters as it compares to a
challenger�s support. We suppose, in particular, that challenger j�s hazard hj;t depends on his
parameters as in Persson and Tabellini 2000, p. 20.25Therefore, in terms of u; a su¢ cient condition is uzzuxx > u2zx.26A su¢ cient condition is uxx < 0:
9
relative support:
hj;t =sSj;tSt
=su(zj;t+1; xj;t+1)
u(zt; xt)(4)
where s is the baseline success hazard (presumably low) of a challenger who is expected to replicate
the incumbent�s policies.27 In other words, s can be interpreted as a measure of incumbency
(dis)advantage. Note that in this �Poisson� political contest, the probability for a challenger to
overthrow the incumbent regime in any interval (t; t+4t) is therefore su(zj;t+1;xj;t+1)u(zt;xt)4t: Hence, our
dynamic model of leadership can be thought of as a continuous sequence of static probabilistic voting
games (Hinich, 1977), which can also be represented by a conditional logit model or qualitative
response model in general.
III.D. Political Equilibrium and the Fundamental Con�ict between Challengers andIncumbent
According to assumption 1, regimes choose public policies only after taking power. Because of
this assumption, when an incumbent chooses his policies, he anticipates what his challengers will
choose once they takeover. Similarly, citizens may also anticipate the new regime�s decisions, but
those decision from regime t will be treated as sunk. This means that rt and xt have no e¤ect on
Vt+1 because those variables are no longer relevant once the t+ 1 regime takes power. Formally,
De�nition 2 (Equilibrium) In any regime fy; w; p; � ; �; i; sg; an equilibrium is an in�nite sequence
fxt; rt; bt; Tt; ct; ht; Vtg1t=0 of policies, number of challengers, success hazard per challenger, andleadership expected value such that:
Vt = maxxt;rt;bt;Tt;ct;ht
rt � �bti+ ctht
such thathtVt+1i+ ctht
� bt � w � 0
ht = su(y � Tt+1; xt+1)u(y � Tt; xt)
Tt = rt + pxt � �y; ct � 0; xt � 0; bt � 0; rt � 0
for all t � 0; with regime t taking as given the sequence fxs; rs; bs; Ts; cs; hs; Vsg from regimes
s � t+ 1:
Note that our de�nition restricts attention to equilibria that are symmetric in the sense that
all challengers to a given regime are identical. Thus, we suppress the j subscript from the per
challenger hazard rate h. Equilibrium public policies are value maximizing for the incumbent, who
anticipates the e¤ect of his policy choices on the number of challengers through the political entry
constraint.28
27The assumption of relative support simpli�es our equilibrium analysis. It is related to binary Luce model(Coughlim 1992, p. 88-92). Moreover, our speci�cation corresponds to the proportional hazards assumption inthe duration analysis literature. In the stationary state equilibrium, our main results are robust to this speci�cation.28A couple of equilibrium concepts have been considered in the patent race literature. One of them is a Nash
10
One important implication of the sequential patent race model comes from the zero pro�t con-
dition, regardless of whether incumbent and challengers are choosing public policies that maximize
their economic value, so it helps to further explore the condition before characterizing the equi-
librium. The zero pro�t condition creates a fundamental con�ict between the incumbent and the
challengers. In particular, using the envelope theorem,
dVtdVt+1
= ���ht
i+ ctht< 0
because the Lagrange multiplier for the entry constraint �� is positive as long as there is a threat
of entry. It is interesting to note that the con�ict exists even when there is no actual challenger in
equilibrium. Because a higher Vt+1 encourages entry, the incumbent has to either increase the size
of entry barriers or reduce rent extraction to keep the number of challengers unchanged � otherwise
the his expected tenure would decline with Vt+1: The incumbent, therefore, wants this continuation
value to be as small as possible.29 It also implies that the e¤ects of permanent determinants of
leadership value of leading are blunted through their impact on challengers. The more challengers
are discouraged, the less the present value of leading is harmed; but the less value is harmed, the less
challengers are discouraged. The easiest way to see this is perhaps in a stationary equilibrium (i.e.
Vt = Vt+1 = V ) with a positive number of challengers, we can substitute the zero pro�t condition
into the incumbent�s value by eliminating ct and obtain a square root relationship between the
present value of leading and the net income �ow from leading:
V =
r(b+ w)(r � �b)
h: (5)
III.E. The Structure of the Political Market: Leviathan, Contestable Political Market,and Entry Equilibria
We de�ne sb, sc, sx; s�; and s� be the critical values of the parameter s that partition the
parameter space according to whether the constraints b � 0, c � 0, x � 0; the entry and the taxconstraints bind, respectively. Three or four market structures are possible in our model, with the
actual outcome depending on the value of the parameter vector f�; s; i; w; � ; p; yg. To eliminatesome trivial cases, we restrict attention to the subset of the parameter space according to the
following assumption.
Assumption 3. (Parameter Restrictions) (a) u�z((1� �)y; p) <1; (b) �w <pu((1��)y;0)ux((1��)y;0) < �y; (c)
px�(y; p) < �y:
equilibrium concept in which incumbents choose their policies taking the number of challengers c as given. Thismight be realistic in some private sector applications, but the government is more commonly modeled as a �leader�that accounts for the e¤ects of its policies on its citizens�behavior. Thus, we treat c as an endogenous variable.29The appendix shows that this fundamental con�ict can create a cycle in the number of challengers because, say,
even-numbered regimes face little entry threat because their few challengers (aspiring to become an odd-numberedregime) expect to face a large entry threat, in turn because their many challengers (aspiring to become an even-numbered regime) expect to face little entry threat.
11
Assumption 3(a) says that the marginal indirect utility is bounded when r = �y so that it allows
for a �leviathan� outcome in which the leadership acts as if it were a pure monopoly: spending
nothing in the public interest. Assumption 3(b) bounds leaders�net cost pu=ux of supplying at
least some strictly positive public spending x.30 Finally, assumption 3(c) simply allows for regimes
that are not constrained by the technology of tax collection, so that not all polities are at the top
of the La¤er curve. We will provide a more detailed discussion of this constraint in section V.C. In
addition to the above parameter restrictions, to simplify our exposition, in what follows, we also
focus our attention on the stationary equilibrium.31
Figure 1 partitions the parameter space according to the market structures that can result
under assumption 3. The number of challengers c is strictly positive above the sc schedule, which
is calculated by inverting the zero pro�t condition facing regime t�s challengers evaluated at ct = 0
and the steady state values ut+1 = u�(y � rt; p) and Vt+1 = rt+�w2psc�; implying sc = i2=�:
The sc schedule is only a rough indicator of political market performance. First of all, public
policies can be quite e¢ cient with c = 0 and quite ine¢ cient with c > 0. Second, a number of model
parameters have a larger e¤ect on b than on c and would therefore would be more readily detected
from data on b than from data on c. For example, when b > 0, the formula for the equilibrium
number of challengers c is:
c = max
(0;
r�
s� i
s
): (6)
Only three of the seven parameters can a¤ect the number of challengers (and only then when the
number of challengers is positive) because the leadership adjusts to parameter changes by changing
entry barriers. For example, the leadership responds to an increase in the opportunity cost of
challenging by reducing the entry barrier b so that there is no net reduction in the number of
challengers. Similarly, an increase in the amount of GDP y that can be taxed has no net e¤ect
on the number of challengers and instead increases the size of political entry barriers. Thus, the
facts that entry barriers are endogenous and practically always positive further enhance the case
for using b rather than c as and indicator of political competitiveness.
Figure 1�s sx schedule provides an economically more interesting indicator of political market
performance. Below this schedule are a variety of �leviathan� outcomes in the sense that the
leadership taxes the maximum and spends none of it in the public interest: the constraint x � 0binds.32 The political market results in leviathans when entry barriers are cheap to maintain (�
small) or when the incumbency advantage is su¢ ciently large even when the leadership is not
blocking entry (i.e., s small or w large). Note that a reduction in the incumbency advantage
30Assuming that it is large relative to the cost of entry barriers � allows for the possibility (but does not require)that entry barriers are sometimes preferable to public spending as a way to enhance political survival. Assumingthat it is small relative to maximum possible rents �y allows for the possibility of a �contestable political market�outcome � in which leaders pay attention to public support even they are facing no actual challengers.31The appendix shows quite similar results for non-stationary equilibria.32Even with an Inada condition (with respect to x ), an �unthreatened leviathan� region is possible because the
leadership may be oblivious to popular support. This is the case considered by the monopoly models of dictatorshipby Olson, Wintrobe, and others
12
can create leadership turnover (c > 0) but does not necessarily motivate spending in the public
interest. In other words, part of the parameter space has a �sequential leviathan� outcome: all
leaders tax the maximum and spend part of their rents erecting entry barriers that extend their
political survival, although not inde�nitely.
Although leadership turnover is not su¢ cient to motivate spending in the public interest, for
some parameter values it does as shown in Figure 1 as the two regions labeled �competitors sup-
plying public spending.�These are the outcomes that would usually be judged to re�ect healthy
competition for public o¢ ce, especially those in which endogenous entry barriers b are small and
an e¢ cient amount is spent in the public interest. Our model also has a public sector analogue to
the �contestable market�hypothesis of Baumol, Panzar, and Willig (1982) and Sutton (1991): a
market may have only one producer but nonetheless perform much like a competitive market. As
we show below, the most e¢ cient outcomes occur in this part of the parameter space.
IV. Economic Causes of Political CompetitivenessEconomic outcomes represented by various parameters in our model � such as military and
communication technologies, standards of living, productivity in the private sector, etc. � have
changed signi�cantly over time, and thereby provide the basis for a theory of democratization. In
this section we give attention to the e¤ects of the parameter w (labor productivity in the private
sector), the parameter y (national income), and the parameter � (the marginal cost of maintaining
and enforcing entry barriers).
IV.A. Complementarity between Rent Extraction and Entry BarriersWe have seen from the previous section that the political market structure (measured by the
number of challengers, and hence regime turnover) in general is a poor indicator of the conduct and
e¢ ciency of the public sector. Moreover, under certain parameter restrictions and hence certain
political market structures (e.g., unthreatened leviathan), political competition is trivial in the
sense that either rent extraction is limited solely by the ability to tax (i.e. r = �y) or entry barriers
are useless (i.e. b = 0). For the remaining but still rich set of political market structures, however,
the degree of political competitiveness (measured by rent extraction or entry barriers) responds
systematically to economic and technological changes. One of the oldest economic doctrines in the
industrial organization literature is that entry barriers give rise to market power, and hence higher
entry barriers should be associated with higher monopoly rent. The following comparative statics
formalize this in our political market setting.
Proposition 1. (Rent Extraction-Entry Barriers Complementarity) Consider any non-leviathanleader who spends a positive amount to deter entry (i.e., rt < �y and bt > 0). Rent extraction and
entry barriers are always complements. Moreover, (a) when the tax constraint is non-binding,
drtdw;dbtdw;drtd�;dbtd�
< 0; (7)
anddrtdy;dbtdy
> 0; (8)
13
and (b) when the tax constraint is binding, (7) is also true, and (8) is true when uzux � uuzx:
The above proposition says that when a leader�s conduct is constrained by political compe-
tition, there is a robust complementary relationship between political rent extraction and entry
barriers. Moreover, this result is independent of political market structure and leaders� ability
to tax. Therefore, even in a contestable political market in which the incumbent leader faces no
actual challengers or in an economy where taxation reaches the top of the La¤er curve, political
competitiveness (measured by the size of entry barriers and rent extraction) will still respond in a
similar way with respect to economic as well as technological changes.
Suppose the number of challengers is positive. When the tax constraint is non-binding, the
incumber�s problem becomes
Vt =1
su(zt+1; xt+1)Vt+1maxxt;bt;rt
(bt + w)(rt � �bt)u(y � rt � pxt; xt): (9)
Holding bt and xt �xed, the optimal rent extraction maximizes the product (rt � �bt)u(y � rt �pxt; xt);which might be interpreted as a social welfare function of net income to leadership and
citizen.33 Less rent extraction is popular and would enhance survival, but the revenue �ow while
in o¢ ce would be smaller. The optimal rent extraction is described by the �rst order condition:
rt � �bt =u(y � rt � pxt; xt)uz(y � rt � pxt; xt)
� u�(y � rt; p)u�z(y � rt; p)
; (10)
because the optimal xt will be chosen to maximize the welfare of the citizen in this case (see section
IV for the details). On the left hand side is the �ow of net income to leadership. On the right
is a function of the citizens� income y � rt: Thus, formula (10) describes the distribution of netincome between citizen and leadership. Note that since u�=u�z is increasing in z by assumption
2(a); the right-hand side of (10) is decreasing in rt; and hence �bt and rt have to move in the same
direction with respect to w and �: In other words, citizens share part of the cost of maintaining
entry barriers.
Although political entry barriers are an important source of political rents, it is constrained by
the cost of maintaining them. The optimal entry barriers satis�es
bt =rt2�� w2: (11)
Note that since y does not appear in the above equation, bt and rt move in the same direction when
y changes. The optimal conditions (10) and (11) imply
��w = rt � 2u�(y � rt; p)u�z(y � rt; p)
:
To show drt=dy and dbt=dy are positive, assume on the contrary that they are not. If drt=dy � 0;33Of course, the citizens�weight in this function is not determined by ethical considerations, but rather by the
e¤ect of their welfare on the regime�s survival via support.
14
rt � 2u�(y�rt;p)u�z(y�rt;p)
will be decreasing in y; which contradicts the left-hand side that is constant in
y: Therefore, we must have drt=dy and hence dbt=dy positive. A similar argument shows thatdrtdw ;
dbtdw ;
drtd� and
dbtd� are negative.
IV.B. Democracies Pay Higher Wages, Political Resource Curse, and The Lipset Hy-pothesis
y is the amount of income available to society (i.e., leaders and citizens combined) and w is
private sector productivity. Holding national income y constant, our model implies that b and w
are negatively correlated.34 Holding w and other parameters constant, an increase in y is associated
with a higher level of b as long as the initial level of b is positive. Because b refers to policies such as
punishments, censorship, and inconveniences created by the incumbent in order to block political
competition, it is a natural indicator of the degree of (non)democracy,35 and our model thereby
o¤ers predictions about correlations between democracy and economic outcomes y and w.
Since w is interpreted as the wage in the private sector, the model predicts that, other things
being equal, more democratic regimes are associated with higher wages. Rodrik (1999) shows that
controlling for income levels and other factors, democracies tend to have higher wages and a higher
labor share. This interesting �nding has been interpreted as evidence that democratic institutions
tend to be more labor-friendly.36 Our theory o¤ers an alternate interpretation: higher wages in
the private sector are a substitute for entry barriers because both discourage challengers, and the
entry barriers are costly for the incumbent to maintain.37
A recession (economic boom) are related cases, in that private sector productivity w is tem-
porarily low (high) without much change in �permanent income.�Thus, our model predicts that
political challengers are less likely during temporary economic booms.38
An interesting and important case in which y increases but w remains constant is a discovery of
natural resource. When the opportunity cost of challenging is held constant, an increase in national
34This is indeed true even when r = �y in equilibrium.35This notion of political competitiveness is consistent with Schumpeter�s (1942) de�nition of democratic method,
which he de�nes as an �institutional arrangement for arriving at political decisions in which individuals acquirethe power to decide by means of a competitive struggle for the people�s vote.� It is also consistent with the way inwhich political scientists empirically measure democracy. For instance, the widely used POLITY IV dataset measuresdemocracy in terms of the competitiveness of political participation and executive recruitment.36 In particular, Rodrik (1999) conjectures that democracies might be more friendly to labor because (1) democratic
regimes are more likely to follow the rule of law, which enhances the bargaining power of labor, (2) democracies areless prone to political instability and discontinuity, which enhances the outside options of employees, (3) democraciesmay directly enhance the bargaining power of labor by allowing freedom of association and of collective bargaining,and (4) the process of political participation and competition may increase the bargaining power and reservationwage of workers by producing legislation and instutitions that are more partial to workers�interests.37Although an economy with higher wages is more democratic in the sense of lower political entry barriers, as
we have seen above, expected regime tenure (or regime turnover) is uncorrelated with wages when b is endogenous.bt + w increasing in w. To see this, assume on the contrary that it is not. The optimal condition for b then impliesdrt=dw < ��; which contradicts the optimal condition for r:38For a temporary reduction in w, holding Vt+1 �xed,
ct =Vt+1bt + w
� i
ht
increases because bt + w decreases as w decreases, and also ht increases with a reduction in w since rt increases andxt decreases. Acemoglu and Robinson�s (2001) model has a similar e¤ect.
15
income increases the potential bene�t of leadership and hence the gain from challenging as well. To
deter entry, the incumbent will run a more oppressive regime and increase his rent extraction, to the
point that the number of challengers remains unchanged in the stationary equilibrium, according
to equation (6).39 This phenomenon is known as the natural resource curse. Tsui (2007) provides
evidence that wealth generated from oil discovery signi�cantly slows down democratic transition.
Economic development is expected to increase both y and w, although whether one increases
proportionally more than another is an empirical question. Our prediction that y and w have
o¤setting e¤ects on b can be compared with the positive cross-country correlation between income
and democracy found empirically. This correlation, known as the Lipset hypothesis (Lipset, 1959),
is one of the central empirical �ndings of comparative political science, and has led some political
scientists as well as economists to interpret democracy as a normal good. However, this view
has been recently challenged by Przeworski and Limongi (1997) and Acemoglu et al. (2006).
Moreover, Haggard and Kaufman (1995) and Acemoglu et al. (2006) show that economic crisis
makes democracy more likely. These empirical challenges contradict the modernization theory,
which asserts a causal link between economic development and democracy. They are, nonetheless,
consistent with our theory, which emphasizes the supply side of democracy. We predict that the
e¤ect of economic development on democracy depends on the source of the development. We have
seen that when income gains are derived from natural resource, the degree of democracy tends
to be weakened. In contrast, when economic development is driven by higher productivity in the
private sector, economic growth may foster democracy because the negative e¤ect of the increase
in the opportunity cost of challenging dominates the positive e¤ect of the increase in national
income on entry barriers.40 Our model says that changes in income and labor productivity alone
(represented by proportional permanent increases in y and w), cannot explain the secular trend of
democratization (represented as an erosion of political entry barriers along with a less concentrated
political market) that occurred in the past two centuries.
IV.C. Enforcement Technological ChangeUnlike income and labor productivity, technological changes a¤ecting the military and other
political entry barriers have not received much attention among economists as determinants of
political transition.41 The parameter � is the marginal enforcement cost of deterring entry, which
may depend on military technology and other technologies available for communication, monitoring
and pursuing criminals, etc. Our theory of political entry barriers predicts that entry barriers b and
rent extraction r fall with �: In other words, technological progress that reduces � would increase
repression and hinder democracy.42 Since entry barriers and rent extraction are complements,
citizens will also su¤er more from more rent extraction. Indeed, when � decreases, not only will b
39For countries with initial b = 0; an increase in y will instead a¤ect the entry margin so that dc=dy > 0:40 In Mulligan and Tsui (2006), we also explore how citizens�freedom demand a¤ects public policy, where freedom
is inversely related to entry barriers. In particular, we show that economic development will foster democracy whenthe income elasticity of freedom demand exceeds its price elasticity in magnitude.41See, however, Huang (2007) for a recent exception.42George Orwell�s famous 1984 raised the possibility that technological progress would favor leadership.
16
increases, enforcement expenditure �b will also rise, because from the optimal condition of b;
d�btd�
=1
2
drtd�
� w2;
which is negative. Moreover, according to equation (6), a reduction in � will have a real negative
impact on the number of challengers. This result is in contrast to changes in w or y; which a¤ect
the size of entry barriers but not the number of challengers.
Emphasizing the incentives of political entry and the cost of entry deterrence, our theory high-
lights a fundamental di¤erence between democratization due to economic factors versus technolog-
ical progress, in which only the latter can simultaneously lead to a decline in entry barriers and an
increase in regime turnover. While being ignored by economists, the role of military technology in
political change has been discussed by political scientists and historians. Finer (1997) argues that
the nature of military technology a¤ects the distribution of military power within a society and
hence the form of government. For example, historically, a disarmed population and a permanent
professionalized force opened a way for an absolutist regime in the Roman, the Byzantine, and the
Chinese empires. Our rent extraction-entry barriers complementarity is also consistent with his
notion of extraction-coercion cycle, which says a dictator can use a military force to extract taxes,
builds up that force, and with it extracts more taxes. Downing (1992) also argues that the shift
from small, decentralized knight service to large standing armies since the late medieval period
in Europe explains the destruction of constitutional government and the rise of military-centered
autocracies.
Communication technologies are often thought to favor challengers to the government.43 The
proliferation of the internet may help challengers to the government to coordinate with each other,
because they no longer have to rely on physical meetings or distribution of hard-copies of their
communications. By raising �; the internet should foster democracy, but not as much as it would if
government policy were held constant, because government increases their enforcement expenditures
in response to the technological change. Examples include attempts by government to censor the
internet or limit its distribution within their borders. Therefore, prior to the 20th century, � may
have increased as military technologies improved. Probably more important in modern times is the
advancement of communications technologies that have the e¤ect of increasing �. Our theory, which
predicts that democracy should decline over time as military technologies developed, and increase
over time as communication technologies improved, is consistent with the secular trends of political
entry barriers and expected regime length. Moreover, it explains why the cross-section and time
series relationships between democracy and development can be di¤erent: oil countries are more
repressive because of pure economic reasons, whereas less developed countries reach democracy
at lower living standards than the developed countries did by importing modern communication
technologies from developed countries.
V. Policy Consequences of Political Competitiveness43 In the famous moon cakes in China, rebels coordinated e¤orts by baking messages inside cakes.
17
When the size of entry barriers is interpreted as the degree of political (non)competitiveness, the
previous section suggests that political regime matters, and in particular political competitiveness is
negatively associated with the extent of rent extraction, although they are indeed jointly determined
in equilibrium. Moreover, we have seen that this relationship does not depend on the tax constraint.
Empirically, we therefore expect correlations between the degree of democracy and public policies
that block political competition or extract rent. In this section, we consider some other consequences
of political competitiveness.
V.A. Does Political Competitiveness A¤ect Social and Economic Policies?Social and economic policies x, like social security, the minimum wage, labor regulation, etc.,
are functionally unrelated to the blocking of political challengers or rent extraction. These policies
a¤ect government�s popular support and thereby its survival as modeled by equation (4). Hence
all regimes will choose the same x, conditional on the income y� r that remains for citizens�publicand private purposes unless spending on x restricts political rents via the tax constraint:
Proposition 2. (Conditional Invariance) Consider any political market structure. (a) When thetax constraint is non-binding (i.e. rt + pxt < �y), the equilibrium social and economic policies
coincide with the citizens�demand (i.e. xt = x�): In particular, conditioning on y � rt and p,
@xt@w
=@xt@�
=@xt@s
=@xt@i
=@xt@�
= 0:
(b) When the tax constraint is binding (i.e. rt + pxt = �y), xt 6= x�, and xt becomes a non-trivialfunction of the regime characteristics. In particular, x will be underprovided.
Regime characteristics a¤ect policies related to the blocking of political challengers and the
amount of political rent, but they do not a¤ect the choice of social and economic policies unless the
tax constraint is binding. The taughtness of the tax constraint provides a necessary and su¢ cient
condition for the invariance result. In particular, among regimes for which the taxation constraint is
non-binding, the equilibrium social and economic policies lie on citizens�desired demand function,
and they will all choose the same policies as a function of citizens� income and �prices� of the
policies. In other words, it is the tax constraint, rather than the lack of political competitiveness
per se, which gives rise to government failure.
To prove part (a) of the proposition, we rewrite the incumbent�s optimization as a Lagrangian:
Vt = maxxt;bt;rt
rt � �bti+ ctht
+ ��(bt + w �htVt+1i+ ctht
) + �� (�y � rt � pxt)
where ht =su(y�rt+1�pxt+1;xt+1)
u(y�rt�pxt;xt) ; ct � 0; bt � 0; and �� � 0 and �� � 0 are the Lagrangian
multipliers for the entry and tax constraints respectively. The policy instrument xt enters in two
ways: as part of the tax constraint and through the challenger hazard ht. If the tax constraint
were not binding, it is simply a question of minimizing the challenger hazard, which is equivalent
to maximizing popular support u(y � rt � xt; xt):44 When the tax constraint is binding, �� > 0;
44A similar argument appears in Persson and Tabellini�s (2000, p. 72) model of politicians�rents under democracy.
18
so that the optimal xt trades o¤ popular support with relaxation of the tax constraint. In other
words, conditional on the same distribution of income, xt will be more expensive to a leader facing
a binding tax constraint, who therefore under-provides xt.
V.B. Does Punishing Dictators Bene�t Citizens?We have built an economic model of regime turnover. One use of such a model is to help predict
the e¤ects of foreign policies intended to encourage �regime change�on the countries targeted. Here
we consider two such foreign policies: reductions in net income of the leadership of the (targeted)
country and the repudiation of odious debts. The e¤ects of these policies are very di¤erent if leaders
maximize tax revenue rather than, as in our model, trade o¤ revenue and survival. Punishing
dictatorial regimes can indeed discourage regime change and have adverse consequences for welfare
of citizens, if the punishment is not designed with attention to political competitiveness and entry
into the market for political leadership.
Suppose a country were to su¤er a perpetual loss L from its net government revenue r � �b,perhaps because it is subject to economic sanctions, has to spend on the military in order to deter
invasion (as the USSR did), etc. If the government had already been extracting the maximum
possible revenue from its citizens, then there is nothing it could do to shift the burden of L to
its citizens. However, only leviathans behave this way. Other leaders in our model � even those
facing no challengers � do not extract the maximum possible revenue from their citizens in order to
enhance regime survival. Consider the case when the number of challengers is positive.45 Leaders
and citizens share the economy�s resources with the citizens according to the income distribution
condition. When the tax constraint is non-binding, the equilibrium conditions become
rt � �bt � L =u�(y � rt; p)u�z(y � rt; p)
and �bt =rt2� �w
2� L2:
As described above, this conditions shows how the equilibrium re�ects a tradeo¤between leadership
and citizen cash �ows. Hence, both citizens and leadership su¤er lost income �ows as a consequence
of the punishment L:
Due to the reduction in leadership income, the marginal bene�t of entry barriers will also be
lower. In equilibrium, rt � L decreasing in L implies dbt=dL < 0 from the optimal condition of
bt: However, it does not follow that punishing dictators helps regime change. To the extent that
challengers expect to su¤er the same loss L if they were to lead the government, the cost of L to
the incumbent is further mitigated. The steady state formulae for V and c imply that the number
of challengers is independent of L; because the leadership value is decreasing in L:
Punishing dictators also reduces xt (i.e. dxt=dL < 0) when xt is a normal good. However, the
tax revenue rt + pxt is increasing in L, since rt increases at a faster rate. If � is not high enough,
the punishment L can induce the government to raise the tax rate to the maximum possible so
that the tax constraint binds and the provision of x becomes ine¢ cient, according to Proposition
2. Thus, if L were intended to help citizens by hastening the demise of the regime, it serves neither
45For other political market structures, see the appendix.
19
purpose: citizens�net incomes are lower, the expected regime tenure remains unchanged, and the
provision of social and economic policy may become ine¢ cient.46
Punishments are more e¤ective, in terms of enhancing freedom and raising citizen utility, when
conditioned on entry barriers.47 For example, if L were known by the leadership to be proportional
to b; the e¤ects of the punishment are isomorphic to the e¤ects of increasing the parameter �:
According to Proposition 1, a higher � is associated with a lower rent extraction and a lower level
of entry barriers. Citizens therefore enjoy higher income than they would under an unconditional
punishment. Moreover, in the steady state equilibrium, such a policy can also encourage regime
change because the number of challengers will also increase with �:
Some of these unintended consequences occur because challengers anticipate a lesser value from
leading. It follows that punishing a dictator hastens his demise if it is known that his successor will
not be punished, regardless of his policies. Perhaps this helps explain why Germany�s unpopular
Weimar republic (unpopular for reasons including the Versailles Treaty and the Great Depression)
made fertile political ground for Hitler�s regime. However, this approach still reduces citizens�
income and, as the German example shows, gives no incentive for the successor regime to be
less oppressive. Lessons like these are familiar in industrial organization, where supply conditions
are given a lot of attention and it is widely recognized that, say, a tax on producers may hurt
competitiveness and consumers more than it hurts producers.
Odious debt � sovereign borrowing for the bene�t of the dictator and not the people � may
o¤er another example. It has been proposed (mostly recently by Jayachandran and Kremer, 2006)
that odious debts be repudiated as a way of hurting dictators and helping their citizens. Studying
all aspects of odious debt repudiation � for example, how to detect odious debt, and to ensure that
all countries can obtain legitimate economic development loans, etc. � is beyond the scope of this
paper, but an economic analysis of odious debt might bene�t from attention to the incentives for
political entry. Suppose, for the sake of argument, that odious debt were known to be accurately
identi�ed and repudiated, so that a market for it would not exist and it would never be issued. In
the context of our model, this means that the dictator cannot borrow to smooth his cash �ows.48
Hence i becomes the dictator�s intertemporal marginal rate of substitution, rather than the world
interest rate, and we presume the former is larger (otherwise he would have no desire to borrow).
In short, the odious debt market collapse can be modeled as an increase in i:
Higher i has no impact on r; b; and x: A regime�s expected tenure increases with i because
46Similarly, when the tax constraint is binding, the optimal conditions imply
��w + L = �y � xt � 2pu((1� �)y; xt)ux((1� �)y; xt)
;
and hence xt is always decreasing in L: The government budget constraint and the optimal condition of bt then implythat rt is increasing in L and bt is decreasing in L: Therefore, even in this case when the citizen�s after-tax incomeremains unchanged, facing a �xed tax budget, the incumbent leader will reduce the provision of xt to increase rt:47Wintrobe (2006, p. 70-1) recommends that foreign policy be conditioned on �repression�or �human rights.�48Legitimate economic development loans are part of the x vector. One Ricardian possibility (see Barro, 1974) is
that the dictator can borrow from his citizens by tilting the time pro�le for r; and then his citizens could borrowfrom abroad. In this case, which we rule out for the sake of argument, odious debt repudiation has no e¤ect on thewelfare of dictator or citizen.
20
the number of challengers decreases.49 Thus, if odious debt repudiation a¤ects incumbent and
challengers equally, it has the unintended e¤ect of lengthening the incumbent�s tenure, while having
no impact on entry barriers or citizen�s income. Intuitively, the incumbent�s bene�ts from b are
in the more distant future than are the bene�ts from r because b serves to lengthen the regime.
If odious debt repudiation a¤ected only the incumbent�s discount rate, then the incumbent would
extract more rent because the higher discount rate makes him less future-oriented.
Proposition 3. (Unintended Consequences of Punishing Dictators) When any non-leviathan
leaders are punished by (a) reducing their political rent by L;
drtdL
2 (0; 1] and dxtdL
< 0:
Although, dbt=dL < 0; dct=dL = 0; as long as the same punishment is anticipated by future leaders,
or (b) increasing their interest rate i,
drtdi=dbtdi=dxtdi
= 0; butdctdi< 0;
or,drtdi> 0;
dbtdi;dxtdi
< 0; anddctdi= 0;
depending on whether challengers are positive, or the political market is contestable, respectively.
V.C. Are Leaders�Conduct Constrained by Political Competition or their Ability toTax?
A fundamental question is whether rents obtained by such regimes are limited by the threat of
challengers or whether they are limited by the taxation constraint (T � �y), because the answer
determines how the regimes� policies will react to economic development, foreign policies that
punish dictators, and other changes in their environment. Figure 2 shows how the answer is related
to market structure. The s� schedule partitions the parameter space according to whether the tax
constraint binds: it binds below and does not bind above.
The distinction between the taxation constraint and the zero pro�t condition can be further
analyzed using the comparative statics for the incumbency advantage parameter s, because it enters
the model only through the zero pro�t condition. This parameter tightens the entry constraint
facing the challenger � because challengers are willing to hurdle higher entry barriers when their
success hazard is higher � with no direct e¤ect on the polity�s tax base. The conceivable range for
s is (0;1), and can be partitioned at four critical s values s�, s� , sc, and either sx or sb dependon the magnitude of �. For s < s�, the tax constraint binds so tightly that x = 0 and r = �y. The
other critical values indicate the largest value of s for which the various non-negativity constraints
bind.50
49Even though the value of leadership involves future cash �ows, incumbent present value (conditional on the valueexpected by successors) does not depend on the discount rate i because i has an exactly o¤setting e¤ect on theforward-looking challengers.50s� is the largest value of s that is consistent with a slack entry constraint. When s the entry constraint is slack,
21
In the leviathan regions, only the technology of tax collection constrains leadership rents, which
is why the leadership spends none of its revenue in the public interest. In the parts of the contestable
market and competitor regions nearest the leviathan regions (below the s� schedule), both entry
and the tax technology constrain leadership rents. Spending in the public interest comes at the
expense of leadership rents, so leaders spend something in the public interest in order to limit entry,
but less than the e¢ cient amount.
In parts of the contestable market and competitor regions above the s� schedule, only entry
constrains leadership rents. The leaders perceive no tradeo¤ between rents and public spending,
so public spending occurs in the e¢ cient amount. In fact, the contestable market region above the
sb and s� schedules has maximum e¢ ciency because: (i) public spending occurs in the e¢ cient
amount (x = x�), (ii) no rents are dissipated by entry barrier maintenance (b = 0), and (iii) no
resources are spent challenging the incumbent ((b+ w)c = 0).
As shown in Figure 2, s� is the smallest of the four critical s values. For the purpose of
illustration, Figure 3 shows the s comparative static for a value of � that is large enough to
encourage public spending for some s, but not so large that entry barriers b are always zero. In
the region of Figure 3 with s 2 (s�; s� ), spending in the public interest px is positive and r < �y.In this region, rent extraction is limited both by competition (or potential of competition) and the
tax constraint. Competition is relevant because it motivates spending in the public interest px,
which, due to the binding tax constraint, reduces the revenue available for rent extraction.
The tax constraint does not bind for s > s� ; rent extraction is limited only by competition,
or the potential for it. In this region, public spending is e¢ cient. Maximum e¢ ciency is achieved
for the smallest s in this region because, in addition to public spending in the e¢ cient amount, no
resources are spent challenging or blocking challengers. For s > sb, leaders dissipate some of their
rents by blocking, or attempting to block, challengers.
Figure 3 is a horizontal slice of Figure 2 at a relatively large value of �. A number of the
comparative statics are similar along horizontal slices at for smaller values of �: more s causes a
monotone reduction in r; monotone increases in x and b, and a hill-shaped response of the number
of challengers c. However, as suggested by Figures 1 and 2, there are also leviathan regions in which
b and/or c are positive while x = 0 and c > 0 regions with a binding tax constraint.51
V.D. The Democracy E¤ect is at Most SmallSuppose for the moment that a �democracy� is a polity with small values of r. Under this
interpretation, part (a) of Proposition 2 allows for an e¤ect of democracy on public policies x
regimes have no reason to spend on x or b because the only purpose of such spending is to discourage entry. In thiscase V = �y=i and the expected pro�t of a challenger is
s�y
i2� w
which would not negative as long as s < s� = wi2
�y.
51Another way is to look at the comparative statics of the tax revenue T when the tax constraint is non-binding.Since z is a normal good, one can show that T is decreasing in �; and hence the tax constraint tends to be non-bindingwhen � is large. Similarly, one can show that T is also decreasing in w; and hence the tax constraint is also less likelyto be binding when w is high.
22
only through an income e¤ect. This income e¤ect is likely small compared to the e¤ects of other
economic and demographic variables determining public policies. First of all, some public policies
may not depend on income per se because they do not reduce citizens�disposable income y�r�px(i.e., p = 0 for some public policies).52 Even for public policies that must be funded by reductions
in citizens�disposable incomes, the income elasticity of citizens�demand for them may be small.53
Finally, even if demand for a public policy were income elastic, the e¤ect of democracy on r may
not be large enough to have a signi�cant proportional e¤ect on y � r.54 Table 1 further illustratesthis point. Each column corresponds to a di¤erent income elasticity of the demand for public
spending. Each row corresponds to a di¤erent assumed impact of democracy of the rent extraction
r, measured as a share of GDP y. The entries in the top panel display the percentage point impact
of democracy, through the income e¤ect, on public spending. The entries in the bottom panel
display the percentage point impact of democracy on the tax rate T=y, assuming that democratic
governments spend 25 percent of GDP.
As shown in the Table�s �rst column (top panel), democracy would have no impact on public
spending if that spending were income inelastic. The impact on the tax rate (bottom panel) is
therefore equal to the impact on the political rent share of GDP. The middle column of the table
shows the unit elastic case, which implies that, holding the price p of xconstant, public expenditure�s
share of GDP is invariant to GDP. In this case, the impact of democracy on public spending ranges
from 1% to 10% for the range of markup impacts shown (more on this range below). The impact on
the tax rate (bottom panel) is smaller in magnitude than the impact on the political rent share by
a factor equal to one minus the fraction of GDP spent by democratic governments. Certainly there
are some categories of public activity that are luxury goods, but it has even been suggested that the
average public activity is a luxury good because total public spending increases more than GDP
over time and across countries.55 When the income elasticity is 1.2 rather than 1.0, the impact on
public spending is slightly larger and the impact on the tax rate somewhat smaller in magnitude.56
52More precisely, when p = 0 and u(z; x) = f(z)g(x); x = argmax g(x): Alternatively, when x is pure economicpolicy that enhances GDP and citizens have no preferences for it directly (i.e. ux � 0), there again will be no incomee¤ect and x will be chosen simply to maximize GDP, regardless of the degree of democracy.53When citizens have quasilinear preferences, the most widely-used utility speci�cation in the political economy
literature (see for example, Persson and Tabillini 2000), the income e¤ect is zero and hence all regimes will choosethe same policy.54Dougan and Snyder (1993), however, argue that tari¤ politics may be an exception, because for some nondemo-
cratic developing countries tari¤ revenue constitutes a signi�cant fraction of government income and hence leadershiprent.55Although Baumol (1967) explains this empirical �nding without a large income elasticity.56An income elasticity of 1.2 �ts the cross-country data pretty well and, because the impact formula is proportional
to the income elasticity, implies proportionally larger percentage impacts on public spending. Mulligan, Gil andSala-i-Martin (2004) regress total government revenue as a percentage of GDP 1973-90 on log real 1960-89 averageGDP per capita and �nd a positive coe¢ cient of 5.2 (see their Table 2), holding some other country charactersitics.At the sample mean taxation percentage of 25.8, the 5.2 coe¢ cient implies an elasticity of 1.2. Persson, Roland, andTabellini (2000, Table 1) report a coe¢ cient closer to zero (and therefore an income elasticity closer to 1.0), basedon a sample of democracies for the years circa 1990. Note that an income elasticity of 1.2 implies that a programthat was 0.5% of GDP in a poor country would be 10% of GDP in a rich country, assuming that the rich and poorcountry had per capita GDP that di¤ered by a factor of 20 and that the other determinants of x demand were heldconstant.
23
The magnitude of the impact of democracy on the rent extraction is also important, although
less studied. Mulligan, Gil and Sala-i-Martin (2004) �nd that the most oppressive regimes only
collect slightly more tax revenue (3% of GDP) than the most democratic regimes (controlling for
development and demographics), which suggests that the middle rows of the table�s to and bottom
panels may be the more relevant ones. (note that the tax rate entries in middle rows of the bottom
range from -5.0 to -2.1). Even if the impact of democracy on the political rent were much larger
� as large as 10% of GDP � the bottom row of the Table�s top panel shows that democracy�s
impact on spending in the public interest is still small compared to some of the impacts that have
been suggested by previous studies.
A number of empirical studies in sociology, economics, and political science have found little
impact of democracy on public policies that probably do not serve as political entry barriers.
For example, Cutright (1965), Jackman (1975), and Pampel and Williamson (1989) observed an
obvious raw correlation between democracy and the introduction of pension and welfare programs,
but pointed out that economic development likely drives social programs, and is correlated with
democracy. Mulligan, Gil, and Sala-i-Martin�s (2004) cross-country study for the years 1960-90 �nds
no signi�cant partial correlation between democracy and the amount of welfare spending, education
spending, the corporate income tax rate, and whether the payroll tax is capped.57 Empirical work
in this area is ongoing; our purpose is to point out how empirical estimates of the democracy�s policy
impacts help gauge the relative importance of entry barriers versus shifts in political in�uence as
de�ning characteristics of democracies
VI. ConclusionsOur paper adapts and extends models of private sector patent races to analyze the economic
causes and policy consequences of political entry. An incumbent leader faces a tradeo¤ between the
magnitude of its �ow of rents and his regime�s survival prospects. Regime survival depends on the
number of challengers and on the policies to be implemented if and when a challenger is ultimately
successful. The number of challengers depends on the probability of a successful challenge, the
present value of rents to be obtained by a successful challenger, the amount of political entry
barriers, and the private sector opportunity costs of political activity. As noted by Sutton (1991) in
the context of private sector entry, these basic incentives are expected to be present under a variety
of institutional arrangements, so that the resulting qualitative comparative statics likely transcend
institutional detail.
One of our results is that competitiveness can be measured in several ways. In our model, the
size b=w of the entry barrier relative to challenger opportunity cost is one measure. Alternatively,
the hazard ch of the incumbent�s losing his job to a challenger is sometimes taken as an indicator of
political competitiveness, for example, when a country is considered �undemocratic�because the
57See also Easterly and Rebelo (1993, p. 436), Lindert (1994), and Mulligan, Gil, and Sala-i-Martin (2002), whofound no cross-country relationship between democracy and a number of government tax and expenditure items.Mulligan, Gil, and Sala-i-Martin (2004) do �nd that democracies are di¤erent in terms of torture, execution, andmilitary policies, but our Proposition 2 does not apply because these policies are closely linked with barriers topolitical entry. For an opposite view, however, see Besley, Persson, and Sturm (2007) and Deacon (2005).
24
incumbent executive seems to have too much electoral success relative to challengers. However, as
with the private sector, measuring competitiveness by ch can be misleading because even an in-
cumbent without challengers may limit his behavior in order to remain that way. Commentators in
the industrial organization literature questioned the Justice Department�s pursuit of private sector
anti-trust cases based on industry concentration.58 Perhaps the same logic applies to the public
sector: our model suggests that regimes with few challengers may nonetheless be quite democratic
in the sense that entry barriers and markup rates are low. Furthermore, the possibility of endoge-
nous political entry barriers blunts the e¤ects of competitiveness on the number of challengers.
Perhaps constitutions or other political institutions designed to �police� the amount of political
competitiveness better serve the public by monitoring entry barriers and markups than monitoring
the number of competitors. In this view, it is appropriate that some of the measures of democracy
developed by political scientists are based on entry barriers rather than the number of competitors.
The �markup� or rent r is commonly used to measure noncompetitiveness in private sector
studies, and it would be interesting to examine such measures for the public sector. However, in
the public sector application, r itself needs not create e¢ ciency costs, whereas public sector entry
barriers have the dual marginal costs of maintenance �b and challenger expenditure cb. In contrast,
potential private sector entry barriers such as advertising, research and development, and vertical
restrictions (e.g., exclusive dealing) are often said to help enhance e¢ ciency. Thus, the main social
costs of noncompetitiveness are fundamentally di¤erent in the private and public sectors.
Measuring competitiveness in one or more of these ways is important because democracies and
nondemocracies have some obvious policy di¤erences including torture, execution, and censorship.
Furthermore, history has plenty of examples of nondemocracies�pursuing reprehensible policies,
but this does not mean that democracies and nondemocracies always, or even usually, have many
di¤erent public policies. The monopolistic competition model suggests that the �product mix and
design�� such as the composition of taxes, spending, and economic regulations � are functions
of economics and demographics, but not regime.59 Although we do not deny that a dictator prefers
more money to less, the fact that his taking is limited by the threat of entry means that he has
an important reason to spend much of the tax revenue in the public interest: it buys him popular
support and thereby regime longevity. Nondemocracies may collect more revenue, but they are not
leviathans.
If we are right that dictators are not leviathans, foreign policies designed to punish them may
have the unintended consequences of postponing regime change and lowering citizens�incomes. The
fact that the value of governing is limited by competition, and not the technology of tax collection,
means that dictators are not 100% marginal claimants on government revenues and pass on their
punishments, at least in part, to citizens and competitors. Punishing dictators conditional on
competitiveness can be more e¤ective in terms of enhancing freedom, and to focus more of the
ultimate incidence of the punishment on the leadership rather than the citizens. Future research
58See Demsetz (1973), Baumol, Panzar, and Willig (1982), Sutton (1991), and Baldwin and Gorecki (1994).59This result therefore also sheds light on the debate on the political origin of Industrial Revolution (North and
Weingast, 1989; Clark 1996, 2007).
25
can use a model like ours to compare and evaluate dictator punishments that are conditioned on b
or c or r or some other measure of political competitiveness.
Our model of dynamic political competition also has implications for recent political events in
oil countries in the wake of the large (and, potentially, largely permanent) oil price increases of
2004-2006. These oil price increases should increase the net incomes of both citizens and leaders
in oil countries which, by itself, encourages challengers. However, many oil countries had already
been among the less democratic countries in the world, and our model predicts further increases
in political entry barriers (and further losses of freedom) as a consequence of the increase in the
value of leading relative to the opportunity cost of challenging. For example, the government of
Iran has been increasingly restricting its citizens�internet access,60 and Venezuela�s government�s
censorship has increased its international visibility. One exception may be Iraq where the value of
leading may someday be quite high due to the country�s oil assets, but the United States and its
allies are attempting to limit �with elections, press freedoms, etc. � political entry barriers. Thus,
regime challengers are twice encouraged in Iraq � once by the expected future value of leadership
and a second time by current-day political freedoms. Perhaps attempts to grab power in Iraq would
have been less intense if the country�s oil assets had not gained so much value since 2003, or entry
into the Iraqi political process were as di¢ cult as in neighboring nondemocratic countries.
60http://www.opennetinitiative.net/studies/iran/
26
VII. AppendixProof of equilibrium and comparative statics
We derive equilibrium comparative statics under various political market structures.
(a) Unthreatened Leviathan: The optimization problem in this case is:
maxx;trt;bt
rt � �bt � Li
subject to the non-negativity constraints and
rt + pxt � �y:
In this case, rt = �y, bt = xt = ct = 0, and hence V = (�y � L)=i: Therefore, the tax constraintis binding, and small changes in w or � will have no e¤ect on any policy. The only e¤ect of an
increase in y is to increase r: Because the entry constraint is non-binding (i.e. htVt+1i � bt�w < 0);in a stationary equilibrium, this requires s < i2w=(�y � L) � s�.(b) Threatened Leviathan: This is the same as the unthreatened leviathan except that the entry
constraint is binding and b > 0. In a stationary equilibrium, the size of entry barriers is determined
by
V =�y � �b� L
iand
sV
i� b� w = 0;
which implies
b =s�y � wi2 � sL
i2 + �sand V =
i(�y + w� � L)i2 + �s
:
While r = �y; which is increasing in y; b is increasing in y and s; but decreasing in w; �; i and L:
Note that to guarantee x = 0; we require that the net marginal bene�t to the incumbent of x
be non-positive, which means that
p � s�V
i
ux((1� �)y; 0)u((1� �)y; 0) =
�s(�y + �w � L)i2 + �s
ux((1� �)y; 0)u((1� �)y; 0) ;
where the equality is satis�ed when s = sx:
(c) Actual Challengers with Non-Binding Tax Constraint : Since c > 0; we can substitute the entry
constraint into the objective function:
Vt = maxxt;bt;rt
(bt + w)(rt � �bt � L)u(y � rt � pxt; xt)su(y � rt+1 � pxt+1; xt+1)Vt+1
When b = 0; the �rst-order conditions are:
rt � �w � L < 0 and rt � L =u(y � rt � pxt; xt)uz(y � rt � pxt; xt)
where xt = argmaxu(y � rr � pxt; xt): First, notice that since the equillibrium policies do not
depend on s; the condition for b = 0 can be written as � > r�Lw ; where r is is the solution to the
27
second equilibrium condition, which is free of s: Because b > 0 was considered in the text, here we
assume that this condition holds.
Assumption 2(a) also implies that drt=dy > 0 and drt=dL > 0:Moreover, since rt is increasing in
L; we must also have drt=dL < 1: Note also since w and � only appear in the inequality constraint,
small changes in w or � will not a¤ect the equilibrium. Moreover, in a stationary equilibrium, the
number of challengers is
c =V
w� i
s=
r(r � L)sw
� i
s
where
V =
rw(r � L)
s
Since, r is increasing in y, c is also increasing in y: Moreover, c is decreasing in w; i and L; but
independent of �: Punishing leaders in this case will reduce regime turnover, because b is �xed.
Finally, since c > 0 in this case, we need to have s > i2wr�L � sc, where r is derived from the above
equilibrium conditions. Note that sc is independent of � when b = 0:
(d) Actual Challengers with Binding Tax Constraint : When the tax constraint is binding (i.e.
rt + pxt = �y) with positive equilibrium x, citizens are consuming a �xed after-taxed income
(1 � �)y and leaders are facing a �xed government budget constraint �y. Therefore, from the
leaders�perspective, there is a direct tradeo¤ between rent extraction and public spending, and
hence assumption 2(a) is neither necessary or su¢ cient for our comparative statics. The optimal
conditions imply
��w = �y � xt � 2pu((1� �)y; xt)ux((1� �)y; xt)
:
Since the right-hand side is decreasing in xt according to assumption 2(b); xt is increasing in � and
w; and hence rt is decreasing in � and w: From the optimal condition of rt; it follows that bt is also
decreasing in � and w:
Finally, di¤erentiating the equilibrium conditions with respect to y; we can show
drtdy
= 2pux� + (1� �) p2uz � (rt � �bt) [(1� �)puzx + �uxx]
3uxp� 2uxx (rt � �bt):
However, pux�+(1� �) p2uz�(rt � �bt) [(1��)puzx+�uxx] > (1� �) p2uz�(rt � �bt) (1��)puzx =(1� �) p2uz � (1� �) puux puzx � 0 when uzux � uuzx: We therefore show that drt=dy is positive.
Even when b > 0, it is still possible to have x = 0. In this case of sequential leviathans,
�pu((1� �)y; 0) + (�y � �b� L)ux((1� �)y; 0) < 0
where
b =�y
2�� w2� L
2�and r = �y.
Hence, b is increase in y; decreasing in w; �; L; and r is increasing in y: The condition for x = 0
28
becomes
� <2pu((1� �)y; 0)wux((1� �)y; 0)
� �y � Lw
;
or equivalently, p > �y+�w�L2
ux((1��)y;0)u((1��)y;0) :
(e) Contestable market with b > 0: First, when the tax constraint is not binding, the problem can
be written as
Vt = maxxt;bt;rt
rt � �bt � Li
such thatsu(y � rt+1 � pxt+1; xt+1)Vt+1
i
1
u(y � rt � pxt; xt)� bt � w = 0
Substituting bt =sVt+1u(y�rt+1�pxt+1;xt+1)
i1
u(y�rt�pxt;xt) �w and xt = x� into the objective function
and maximizing with respect to r, the �rst-order condition is:
1
i� �i
sVt+1u(y � rt+1 � pxt+1; xt+1)i
uzu2= 0:
In a stationary equilibrium, Vt+1 = Vt and ht = s; and hence
V =i (r + �w � L)
i2 + �s:
Substituting back to the �rst-order conditions and rearranging, in stationary equilibrium we have
u�
u�z� �s(r + �w � L)
i2 + �s= 0:
Assume that r is decreasing in y. Assumption 2(a) imples that u�=u�z increases as y increases,
but �s(r+�w�L)i2+�s
will become smaller when y increases, which contradicts the equilibrium condition.
Therefore, we must have dr=dy > 0. This also implies dV=dy > 0: Indeed, similar arguments show
dr=dy 2 (0; 1), dr=dw < 0, dr=d� < 0; dr=di > 0, and dr=ds < 0:Assume that r is decreasing in L. From the above equation, u�=u�z increases as L increases,
but �s(r+�w�L)i2+�s
will become smaller when L increases, which contradicts the equilibrium condition.
Therefore, we must have dr=dL > 0. Since r is increasing in L; u�=u�z decreases as L increases, we
must have r � L decreases in L from the above equation.
Substituting V into the zero pro�t condition, in a stationary equilibrium, we obtain:
b =s(r � L)� wi2
i2 + �s:
It follows that b and r are complements in w; �;and y. Moreover, b is decreasing in i and L:
Unlike the case when c > 0; tax revenue r + px is decreasing in s in a contestable market.
29
Next, consider the case when the tax constraint is binding and x > 0: The problem becomes
maxxt;rt
�y � pxt � Li
� �i(sVt+1u(y � rt+1 � pxt+1; xt+1)
i
1
u((1� �)y; xt)� w):
The equilibrium formula V = i(r+�w�L)i2+�s
together with the �rst order condition with respect to x
implies�s (�y + �w � L)
i2 + �s=pu((1� �)y; x)ux(1� �)y; x)
:
Since the left-hand side is increasing in w and the right-hand side in increasing in x; dx=dw >
0;and hence dr=dw < 0 from the tax constraint. Similar arguments imply dx=d� > 0; dx=dL < 0,
dx=di < 0; and hence dr=d� < 0; dr=dL > 0 and dr=di > 0:
Finally, di¤erentiating with respect to y;we obtain
dx
dy=ux�s� � puz (1� �)
�i2 + �s
�+ (1� �)�suzx (r + w� � L)
pux (i2 + 2�s)� �suxx (r + w� � L)
and hence
dr
dy=(p�ux + p
2 (1� �)uz)�i2 + �s
�� (�s�uxx + ps�uzx (1� �)) (r + w� � L)
pux (i2 + 2�s)� �suxx (r + w� � L):
Note that (r + w� � L) = p(i2+�s)u�sux
;and hence dr=dy > 0 when uzux � uuzx:The e¤ect on b is similar to the case when the tax constraint is non-binding.
(f) Contestable market with b = 0 : First, when the tax constraint is non-binding, the problem can
be written as
Vt = maxxt;rt
rt � Li
+ �(w � su(y � rt+1 � pxt+1; xt+1)Vt+1i
1
u(y � rt � pxt; xt))
The �rst-order conditions are:
1
i� �su(y � rt+1 � pxt+1; xt+1)Vt+1
i
uzu2= 0
andsu(y � rt+1 � pxt+1; xt+1)Vt+1
i
1
u(y � rt � pxt; xt)= w
with xt = argmaxu(y � rt � pxt; xt):In a stationary equilibrium,
V =iw
sand r =
i2w
s+ L
and hence r is increasing in w; i and L, decreasing in s; but independent of y and �:
30
Note that to make sure b = 0 is optimal, we need � > i�; and hence
� �u(y � i2w
s � L� px; x)wuz(y � i2w
s � L� px; x)
with equality when s = sb:
Finally, since the tax constraint is non-binding, we must have px + r = px + i2ws + L < �y;
where x = argmaxu(y � i2ws � L � px; x); which is increasing in s and decreasing in w; i; and L:
Since r is decreasing in s and x is increasing in s; tax revenue is decreasing in s when z is a normal
good. As s decreases, T increases and the tax constraint becomes binding at s� = i2w�y�px�L ; which
does not depend on �:
Next, suppose the tax constraint is binding. The problem becomes
Vt = maxxt
�y � px� Li
+ �(w � su(y � rt+1 � pxt+1; xt+1)Vt+1i
1
u((1� �)y; xt)):
In a stationary equilibrium, V = iws and r =
i2ws + L; and hence from the tax constraint and the
�rst order condition,
x =�y � Lp
� i2w
sp;
which is increasing in s; � ; y and decreasing in i; w;and L: Therefore, in a contestable political
market with b = 0; the taughtness of the tax constraint will a¤ect the economic and social policy
only but will have no e¤ect on rent extraction.
Finally, note that to ensure b = 0; we need � > i� and hence
� �pu((1� �)y; �y�Lp � i2w
sp )
wux((1� �)y; �y�Lp � i2wsp )
;
with equality when s = sb:�
Almost Stationary Equilibria and Political Cycles
Our model has a unique stationary equilibrium. For s > i2=�, the stationary equilibrium
number of challengers cand regime values V have square root closed form solutions:
V =
r�
s(b+ w) and c =
r�
s� i
s:
Non-stationary equilibria also exist. Interestingly, if s > i2=�, these equilibria are almost
stationary in the sense that most of the variables are constant over time and have the same values
as in the stationary equilibrium. Only regime values and the number of challengers are vary over
time, and these follow a every-other-regime cycle. Regime net income �ows are the same as in
the stationary equilibrium, but regime values cycle every other regime because of the cycle in the
number of challengers.
31
To see this, notice that the zero pro�t condition holds with equality for an equilibrium su¢ ciently
near the stationary equilibrium (because s > i2=�). As shown for the proof of Proposition 1, this
implies:
Vt =1
su(zt+1; xt+1)Vt+1maxxt;bt;rt
f(bt + w)(rt � �bt)u(y � rt � (p� q)xt; xt)g
The parameters of the maximization problem in square brackets are the same for every regime,
which means that all regimes choose the same policies as in the stationary equilibrium. Citizens
therefore enjoy the same constant utility �ow, and leaders the same net income �ow, as in the
stationary equilibrium. However, this does not imply that Vtnecessarily equals Vt+1. Rather, we
have
V0V1 = V1V2 = V2V3 = V3V4:::
In other words, V can alternate between two values � one value Veven for even numbered
regimes and another value Vodd for odd number regimes. Because b and h are constant over time,
the zero pro�t condition implies that the number of challengers alternates from one regime to the
next, with more challengers occurring prior to the high value regimes.
There are a continuum of these non-stationary equilibria, half with few challengers for the
initial regime and many challengers for his successor. For these equilibria, the initial regime has
a longer expected lifetime and therefore a higher expected value than his successor. One of the
non-stationary equilibria has no challengers for the initial regime, which means that the initial
regime lasts forever. This is a more dramatic example of the political contestable market hypothesis
because this unchallenged regime has public policies and income �ows just like those for an otherwise
identical polity that does have challengers and turnover.�
32
References
[1] Acemoglu, Daron, Simon Johnson, James A. Robinson, and Pierre Yared. �Income and Democ-
racy.�manuscript, MIT, October 2006
[2] Acemoglu, Daron and James A. Robinson. �Why Did the West Extend the Franchise? Democ-
racy, Inequality, and Growth in Historical Perspective.� Quarterly Journal of Economics.
115(4), November 2000: 1167-99.
[3] Acemoglu, Daron and James A. Robinson. �A Theory of Political Transitions.� American
Economic Review. 91(4), September 2001: 938-63.
[4] Acemoglu, Daron and James A. Robinson. Economic Origins of Dictatorship and Democracy.
Cambridge, MA: Cambridge University Press, 2006.
[5] Barro, Robert J. �Are Government Bonds Net Wealth?.�Journal of Political Economy. 82(6),
November 1974: 1095-1117.
[6] Baumol William J. �Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis.�
American Economic Review 57(3), June 1967: 415-426
[7] Baumol,William J. John C. Panzar, and Robert D. Willig. Contestable Markets and the Theory
of Industry Structure. New York: Harcourt Brace Jovanovich, 1982.
[8] Baye, Michael R. and Heidrum C. Hoppe. �The Strategic Equivalence of Rent-Seeking, Inno-
vation, and Patent-Race Games.�Games and Economic Behavior. 44(2), 2003: 217-226.
[9] Becker, Gary S. �Competition and Democracy.� Journal of Law and Economics. 1(3), 1958:
105-9.
[10] Besley, Timothy and Stephen Coate. �An Economic Model of Representative Democracy.�
Quarterly Journal of Economics. 112(1), February 1997: 85-114.
[11] Besley, Timothy and Stephen Coate. �Sources of Ine¢ ciency in a Representative Democracy:
A Dynamic Analysis.�American Economic Review. 88(1), March 1998: 139-156.
[12] Besley, Timothy, Torsten Persson, and Daniel M. Sturm. �Political Competition, Policy and
Growth: Theory and Evidence from the United States.� NBER Working Paper no. 11484,
CEPR Discussion Paper no. 5138.
[13] Boix, Charles. Democracy and Redistribution. Cambridge MA: Cambridge University Press,
2003.
[14] Brennan, H. Geo¤rey and James. M. Buchanan. The Power to Tax: Analytical Foundations
of a Fiscal Constitution. Cambridge, MA: Cambridge University Press, 1980.
33
[15] Breton, Albert. The Economic Theory of Representative Government. Chicago: Aldine Pub.
Co., 1974.
[16] Bueno de Mesquita, Bruce, Alastair Smith, Randolph M. Siverson, and James D. Morrow. The
Logic of Political Survival. Cambridge, MA: MIT Press, 2003.
[17] Clark, Gregory. �The Political Foundations of Modern Economic Growth: England, 1540-1
800.�Journal of Interdisciplinary History. 26(4), 1996: 563-587.
[18] Clark, Gregory. A Farewell to Alms: A Brief Economic History of the World. Princeton Uni-
versity Press, 2007.
[19] Coughlin, Peter J. Probabilistic Voting Theory. Cambridge, MA: Cambridge University Press,
1992.
[20] Crain, W. Mark. �On the Structure and Stability of Political Markets.� Journal of Political
Economy. 85(4), August.1977: 829-842.
[21] Cutright, Philips. �Political Structure, Economic Development and National Social Security
Programs.�American Journal of Sociology. 70(5), March 1965: 539-55.
[22] Davis, Steven J., Kevin M. Murphy and Robert H. Topel. �Entry, Pricing and Product Design
in an Initially Monopolized Market.� Journal of Political Economy. 112(1), February 2004:
S188-225.
[23] Deacon, Robert T. �Dictatorship, Democracy and the Provision of Public Goods.�University
of California at Santa Barbara, Economics Working Paper Series, no. 1172, 2005.
[24] Dixit, Avinash and Carl Shapiro. �Entry Dynamics with Mixed Strategies.� in Thomas III,
Lacy.G. (Ed.), The Economics of Strategic Planning: Essays in Honor of Joel Dean. Lexington
Books, Lexington, MA, 1986: 63-79.
[25] Dougan, William R. and James M. Snyder. �Are Rents Fully Dissipated?� Public Choice.
77(4),December 1993: 793-813.
[26] Downing, Brian M. The Military Revolution and Political Change. Princeton University Press,
1992.
[27] Easterly, William and Sergio Rebelo. �Fiscal Policy and Economic Growth: An Empirical
Investigation.�Journal of Monetary Economics. 32(2), December 1993: 417-58.
[28] Feddersen, Timothy J., Itai Sened, and Stephen G. Wright. �Rational Voting and Candidate
Entry under Plurality Rule.�American Journal of Political Science. 34(4), November 1990:
1005-1016.
[29] Finer, Samuel E. The History of Government. Oxford University Press, 1997.
34
[30] Friedman, Daniel and Donald Wittman �Why Voters Vote for Incumbents But Against In-
cumbency: A Rational Choice Explanation.�Journal of Public Economics. 57(1), May 1995:
67-83.
[31] Gallego, Maria and Carolyn Pitchik. �An Economic Theory of Leadership Turnover.�Journal
of Public Economics. 88(12), December 2004: 2361-2382.
[32] Gelman, Andrew and Gary King. �Estimating Incumbency Advantage without Bias.�Ameri-
can Journal of Political Science. 34(4), November 1990: 1142-1164.
[33] Goolsbee, Austan and Chad Syverson. �How Do Incumbents Respond to the Threat of Entry?
Evidence from the Major Airlines.�manuscript, University of Chicago, November 2006.
[34] Grossman, Herschel I. �A General Equilibrium Model of Insurrections.�American Economic
Review. 81(4), September 1991: 912-921.
[35] Grossman, Herschel I. �The State: Agent or Proprietor?� Journal Economics of Governance.
1(1), March 2000: 3-11.
[36] Grossman, Herschel I. and Suk Jae Noh. �A Theory of Kleptocracy with Probabilistic Survival
and Reputation.�Economics and Politics. 2(2), July 1990: 157-171.
[37] Grossman, Herschel I. and Suk Jae Noh. �Proprietary Public Finance and Economic Welfare.�
Journal of Public Economics. 53(2), February 1994: 187-204.
[38] Haggard, Stephan and Robert R. Kaufman. The Political Economy of Democratic Transitions.
Princeton University Press, 1995.
[39] Hinich, Melvin J. �Equilibrium in Spatial Voting: The Median Voter Result is an Artifact.�
Journal of Economic Theory. 16(2), December 1977: 208-19.
[40] Huang, Fali. �The Coevolution of Economic and Political Development from Monarchy to
Democracy.�Manuscript, Singapore Management University, November 2007.
[41] Jackman, Robert W. Politics and Social Equality: a Comparative Analysis. New York: John
Wiley, 1975.
[42] Jayachandran, Seema and Michael Kremer. �Odious Debt.�American Economic Review. 96(1),
January 2006: 82-92.
[43] Lee, Tom and Louis L. Wilde. �Market Structure and Innovation: A Reformulation.�Quarterly
Journal of Economics. 94(2), March 1980: 429-436.
[44] Lindert, Peter. �The Rise of Social Spending, 1880-1930.�Explorations in Economic History.
31(1), January 1994: 1-37.
35
[45] Lipset, Seymour M. �Some Social Requisites of Democracy: Economic Development and Po-
litical Legitimacy.�American Political Science Review. 53(1), March 1959: 69-105.
[46] Lizzeri, Alessandro and Nicola Persico. �The Provision of Public Goods under Alternative
Electoral Incentives.�American Economic Review. 91(1). March 2001: 225-239.
[47] Lott, John R., Jr. �Brand Names and Barriers to Entry in Political Markets.�Public Choice.
51(1), January 1986: 87-92.
[48] Lucas, Robert E. Jr. �On the Size Distribution of Business Firms.�Bell Journal of Economics.
9(2) Autumn, 1978: 508-523.
[49] Mulligun, Casey B., Ricard Gil, and Xavier Sala-i-Martin. �Social Security and Democracy.�
NBER Working paper no. 8958, May 2002.
[50] Mulligun, Casey B., Ricard Gil, and Xavier Sala-i-Martin. �Do Democracies Have Di¤erent
Public Policies than Nondemocracies?� Journal of Economic Perspectives. 18(1), Winter 2004:
51-74.
[51] Mulligan, Casey B. and Xavier Sala-i-Martin. �Gerontracy, Retirement, and Social Security,�
NBER Working Papers no. 7117, 1999.
[52] Mulligan, Casey B. and Kevin K. Tsui. �Political Competitiveness.�NBER Working Papers
no. 12653, October 2006.
[53] Myerson Roger B. �E¤ectiveness of Electoral Systems for Reducing Government Corruption:
A Game-Theoretic Analysis.�Games and Economic Behavior. 5(1), 1993: 118-132.
[54] Myerson Roger B. �Theoretical Comparisons of Electoral Systems � Theory and Applica-
tions.�European Economic Review. 43(4), April 1999: 671-697.
[55] Niskanen, William A. �Autocratic, Democratic, and Optimal Government.�Economic Inquiry.
35(3), July 1997: 464-79.
[56] North, Douglass C. Structure and Change in Economic History. W. W. Norton & Company,
1981.
[57] North, Douglass C, John Joseph Wallis, and Barry R. Weingast.�A Conceptual Framework for
Interpreting Recorded Human History.�NBER working paper series, no. w12795, December,
2006.
[58] North, Douglass C. and Barry R. Weingast. �Constitutions and Commitment: The Evolution of
Institutions Governing Public Choice in Seventeenth-Century England.�Journal of Economic
History. 49(4), 1989: 803-832.
[59] Olson, Mancur. �Dictatorship, Democracy, and Development.� American Political Science
Review. 87(3), September 1993: 567-76.
36
[60] Olson, Mancur and Martin McGuire. �The Economics of Autocracy and Majority Rule: The
Invisible Hand and the Use of Force.� Journal of Economic Literature. 34(1), March 1996:
72-97.
[61] Orwell, George. 1984. New York, Signet Classics, 1950.
[62] Pampel, Fred C. and John B. Williamson. Age, Class, Politics and the Welfare State. Cam-
bridge, MA: Cambridge University Press, 1989.
[63] Polo, Michele. �Electoral Competition and Political Rent.�Innocenzo Gasparini Institute for
Economic Research, Milian Working Paper No. 144, December 1998.
[64] Persson, Torsten, Gerard Roland, Guido Tabellini. �Comparative Politics and Public Finance.�
Journal of Political Economy. 108(6), 2000: 1121-1161.
[65] Persson, Torsten and Guido Tabellini. �The Size and Scope of Government: Comparative
Politics with Rational Politicians.� 1998 Marshall Lecture, in European Economic Review.
43(4-6), 1999: 699-735.
[66] Persson, Torsten and Guido Tabellini. Political Economics: Explaining Economic Policy. Cam-
bridge, MA: MIT Press. 2000.
[67] Persson, Torsten and Guido Tabellini. The Economic E¤ects of Constitutions. Munich Lectures
in Economics. Cambridge, MA: MIT Press. 2003.
[68] Polity IV Project. Political Regime Characteristics and Transition, 1800-2000. Electronic data
�le (version p4v2000). College Park, MD: CIDCM, University of Maryland, 2000.
[69] Przeworski, Adam and Fernando Limongi. �Modernization: Theories and Facts.�World Poli-
tics. 49, 1997: 155-183.
[70] Reinganum, Jennifer F. �Innovation and Industry Evolution.�Quarterly Journal of Economics.
100(1), February 1985: 81-99.
[71] Rodrik, Dani. �Democracies Pay Higher Wages.� Quarterly Journal of Economics. 114(3),
August 1999: 707-738.
[72] Schumpeter, Joseph A. Capitalism, Socialism, and Democracy. New York: Harper and Broth-
ers, 1942.
[73] Stephen, James Fitzjames. Liberty, Equality, Fraternity. London: Smith, Elder, & Co., 13
Waterloo Place, 1873.
[74] Stigler, George J. �Economic Competition and Political Competition.�Public Choice. 13, Fall
1972: 91-106.
[75] Sutton, John. Sunk Cost and Market Structure. Cambridge, MA: MIT Press, 1991.
37
[76] Svolik, Milan. �A Theory of Government Dynamics in Authoritarian Regimes.�manuscript,
University of Illinois at Urbana-Champaign, April 2006.
[77] Tsui, Kevin K. �More Oil, Less Democracy: Evidence from Worldwide Crude Oil Discoveries.�
manuscript, Clemson University, June 2007.
[78] Tullock, Gordon. �Entry Barriers in Politics.�American Economic Review. 55(1), March 1965:
458-466.
[79] Tullock, Gordon. Autocracy. Boston, MA: Kluwer Academic Publishers, 1987.
[80] Wintrobe, Ronald. �The Tinpot and the Totalitarian: An Economic Theory of Dictatorship.�
American Political Science Review. 84(3), September 1990: 849-72.
[81] Wintrobe, Ronald. The Political Economy of Dictatorship. Cambridge, MA: Cambridge Uni-
versity Press, 2000.
[82] Wittman, Donald A. �Why Democracies Produce E¢ cient Results?� Journal of Political
Economy, 97(6), December 1989: 1395-1424.
[83] Wittman, Donald A. The Myth of Democratic Failure: Why Political Institutions are E¢ cient.
Chicago: University of Chicago Press, 1995.
[84] Wohlgemuth, Michael. �Entry Barriers in Politics, or: Why Politics, Like Natural Monopoly,
Is Not Organised as an Ongoing Market-Process.�Review of Austrian Economics. 12(2), 1999:
175-200.
38
Figure 1. Market Structures Consistent with Stationary Equilibria
β
The Figure illustrates comparative statics with respect to s and β, the challenger hazard and price of entry barriers, respectively. Three steady state outcomes are examined: entry barriers b, public policy x, and the number of challengers c. The parameter space is divided into seven regions according to which (if any) of the non-negative constraints bind: 3 leviathan regions (no spending in the public interest), 2 contestable market regions (no challengers, but spending in the public interest), and 2 regions with competitors supplying public spending.
0 ssπ= i2w/(τy)
τy/w
unthreatened leviathan
threatened leviathan
competitors supplying public
spending
c > 0
c = 0
c = 0
c = 0
c > 0
c > 0
c = 0sc= i2/β
contestable market
sb
sequential leviathansx = 0
x > 0
x = 0
x > 0
x > 0
x > 0
x = 0
b > 0
b = 0
b = 0
b > 0
b > 0
b = 0
b > 0
sx
Figure 2. Entry vs. the Laffer Curve as Constraints on Leadership Rents
β
The Figure illustrates comparative statics with respect to s and β, the challenger hazard and price of entry barriers, respectively. The parameter space is divided into regions according to whether the entry constraint binds, the tax constraint binds, or both.
0 ssπ= i2w/(τy)
τy/w
sc= i2/β
sb
efficient outcomes
only entry constrains
leadership rents
both constrain leadership rents
only Laffer curve constrains leadership rents
sx
sτ
Figure 3. Taughtness of the Entry Constraint (steady state)
r, x, b, c r = τy (leviathan region)
The Figure illustrates comparative statics with respect to s, the challenger hazard facing current and all future regimes. Three steady state outcomes are examined: rent extraction r, public policy x, and the number of challengers c. “Efficient” public policy x* is shown for reference.
0 ssπ= i2w/(τy) sτ
c
x = x*
r
x*
x
sc= i2/β
contestable market region
sb
b
fully efficient
dr/y 0 1 1.2
-0.01 0 1.0 1.2-0.03 0 3.0 3.7-0.05 0 5.1 6.2-0.10 0 10.5 12.6
-0.01 -1.0 -0.8 -0.7-0.03 -3.0 -2.3 -2.1-0.05 -5.0 -3.8 -3.5-0.10 -10.0 -7.5 -7.0
Panel A entries display 100*(ln x for democracy - ln x for nondemocracy)Panel B entries display 100*(T /y for democracy - T /y for nondemocracy)Democracy is assumed to have r = 0. Nondemocratic r exceeds democratic r by dr
Table 1: Effect of democracy on public spending and taxation via the income effect
Income elasticity of demand for public spending
Panel A: Percentage effect on public spending amount
Panel B: Percentage point effect on the tax rate , assuming democratic T /y = 0.25